Hi
A few weeks ago I posted on making a non immersed level dector, using capacitance.
I could have purchased a metal detector, but foolishly with my (very) limited knowledge, I decided to have a go at making my own. Well, what a nightmare,
The story so far.( im treading a well worn path as apparently Amps oscillate- oscillators don’t)
The difficulties started with the sense oscillator, this is usually done with a colpitts,unfortunately this relies on a split cap which I didn’t really want, so, I decided to go for a Hartley and split the coil.
Secondly the capacitance on the actual plate sensor is only about 5 to10 pf, so the a tuning inductor would need to be around 100mh at the few kilohertz these things normally operate at, unfortunately the reactance impedance would be horrendous at low frequencies, also there always seems to be a residual 10p of parasitic,
Fortunately there dosent seem to be any set frequency for this kind of thing, apart from I suppose, keeping out of the broadcast spectrum, so up to a few hundred khertz should be ok, and the high value inductors are readily available from Cricklewood.
Tried a couple of bfo circuits,
Tried a 555 using the sense cap as part of the frequency timer, this didn’t work.
(Dosent have to be a sine wave)
I then realized that the conventional detector type circuit had a fatal flaw in that they rely on the user making constant adjustment each use to compensate for drift, temperature, stray capacitance etc, this would need to be be a permanant fixture

So back to the Hartley,see pic. I had a 10 meg xtal handy, put this in the circuit and amazingly (for me) it works, I then hoped that in operation, any variation in the tuned circuit capacitance would kill the oscillation and trigger an output stage.
So now ive got a nice steady 10 mhz output from the oscillator,with a 9 to 11 mhz tunable tuned circuit (a rectifier/ comparator final output stage that triggers at .8 volts, ready to go.)snag is , I would have expected the oscillations to cease when I significantly vary the tuned circuit, away from the xtal frequency, unfortunately the output frequency moves from 10 mhz and just continues to happily follow the tuned circuit frequency right down to 7 mhz .
Its possible that the amplitude drops, though, that would still work, but I don’t have an oscilloscope, and ive found simulators (mine anyway) struggle with oscillators over a few khz and tricky to start.
Interestingly, in real life they rely on noise to initiate which I suppose is difficult to simulate.
Im not too concerned about the outcome, this is proving to be a fascinating and enjoyable subject in its own right.

Any and all thoughts most welcome.


im a bit puzzled on beat frequency oscillators.when they mix they cancel each other out. how? as there is no mechanism to set the phase position.


When I next have a play, I think I will leave the hartley as it is and put a LC tank from output to ground on the output and use this as the sense capacitor.

Regards

dougal

 

Many of the Picaxe uC's have a "touch" command, it is a capacitive sensor. I believe it works by upsetting a local oscillator and converting that into a digital word.

 

Would this sensor work for your application?

hgmjr

 

hi
thank you for your replies,
jaguarjoe, yes you are quite right, i shall probably end up with a local oscillator/ sense oscillator pair. however in the interest of simplicity and cost, especially using xtal/s i was hoping to do it using a single oscillator,hoping to finish that tonight or tomorrow and get back.
thanks HgMJR i missed that one, i am hoping though to have the sensor on the outside,as i have no access from the top.i'll try and get a pic of the header tank.

oops, i forgot to edit the pic- the xtal should of course read 10 mhz,
the pb switch i should have deleted, was a vain attempt to start oscillations on the sim

thanks again

dougal

 

hi again
finally got round to doing a bit more,it seems to work well on the bench,(a bit surprising as ive tried 3 different circuits without success) still plenty to do though,
bits of 3 circuits so plenty of tidying up, also the hartley coils can be changed to a single mini inductor now im not using it as the sensor,
the secret seems to be running it at far higher frequency than normal ,so before i go any further could someone please help on this,should i stick with 10 mhz?
i know theres a certain military man who does these things with his eyes closed,
pic- i havent finished putting in the values yet ,
regards
dougal

 

I would go for a less complex solution. See the attached.

The 4" square plates that make up C1 can be conformal coated (clear acrylic lacquer spray paint would work fine) so that there is no actual contact between the plates and the water.

 

thanks Wookie, that looks interesting,i have managed to get a pic of the tank, there is no real access from the top, and the existing sender is well down the tank.
if nothing else ive made a good alarm circuit- it will detect my finger at 3 inches.and ive learnt a lot.

i did say it was a mess, i hate breadboards so i tend to knock up a pcb, only takes a few minutes, and leave some length on the componants in case i need to substitute.and tidy up later, probably explains some of the strays though.
i still havent decided on the frequency, a 32khz xtal might be too low.


regards
dougal

other pics the plate sensor, (had to use a blue led),i can lose most of whats in the pic.